Self-Report Outcome Measures for Adult Hearing Aid Services: Some Uses, Users, and Options

Introduction

In many instances throughout the world, audiology developed as a technically orientated discipline with heavy emphasis on the engineering aspects of equipment and hearing aids. Many of the techniques in audiology were primarily diagnostic in origin and the rehabilitative elements of hearing aid provision, beyond simple amplification, were relatively slow to develop. I distinctly remember as a young scientist, on entering the field of hearing and its assessment, encountering articles whose primary purpose was to establish and justify the very position and role of self-report measures in the scientifically oriented practice of audiology. Such arguments were necessary to bring to the attention of both researchers and practitioners the primary focus of intervention for hearing deficits beyond the fitting of simple amplification.

Few would attempt to argue that the primary role of a rehabilitative (as opposed to a diagnostic) audiologist or service is to fit hearing aids. Rather there is an almost universal acceptance that the objective of interventions is to overcome the deficits in the domains of disability, handicap, and health-related quality of life that occur as a consequence of impaired hearing. For definition of the terms disability and handicap and their likely successors activity and participation, readers are referred to publications provided by the World Health Organization (WHO, 1980; WHO, 1997). The experience of disability and handicap can only be understood and assessed via reports from impaired individuals and those with whom they interact. Inspection of any modern audiologic journal or textbook will demonstrate evaluations that revolve around the reports from hearing-impaired listeners of the impacts of hearing impairment and the extent to which interventions and improvements in interventions either do or do not alleviate those deficits. Self-report measures of disability and handicap, and hence the benefits of intervention, now constitute an unassailable element in our understanding of the consequences of impaired hearing and essential components for the evaluation and optimization of interventions designed to alleviate those deficits. Exclusive reliance on procedures such as aided benefit scores on speech tasks has all but disappeared. Self-report has come of age.

It is fair, however, to point out that in everyday audiologic practice, the formalized use of self-report measurements to evaluate and guide management options has not kept pace with their standing in more research and evaluation-orientated exercises. The objective of this article is to examine some of the applications, constraints, and requirements for and upon self-report outcome measures. These mainly derive from research enterprises, but are directed toward routine clinical environments and the peculiar constraints and restrictions that they impose. Given the wide literature concerning self-report measures, the article will focus on a series of issues rather than attempt a general survey of the research and clinical literature. I do intend, however, to provide appropriate references for wider reading so that the particular arguments and viewpoints that I advance can be evaluated in the context of a wider knowledge base.

Potential Uses and Users

It is all very well for audiologic researchers and practitioners to promote the use of self-report measures of outcome in a rather nonspecific manner. However, this generalized promotion may be self-defeating if we cannot point to the specific uses to which they might be put and the specific advantages to their use by practitioners and clients. I attempt to address these issues in the context of both the use to which the information might be put and the users who might avail themselves of the particular information.

All audiologists, be they orientated toward research or clinical practice, are aware of the explosion of technologic and rehabilitative options that have become available. Although this may be facilitated by a particular type of technology (e.g., digital or digitally programmable) it is not really an issue concerning the technological implementation. Rather we now have at our disposal a whole host of hearing aid features and their fitting options, and consequently we are faced with the task of selecting features of a hearing aid fitting and matching them to the needs and desires of hearing-impaired people to overcome their disabilities and handicaps. I will not attempt to provide an exhaustive review of the features available, as these will inevitably become out of date as new signal processing techniques are implemented and the research community develops innovative rehabilitative options. Some of the features may reside in the restricted technologic domain (e.g., directional microphones, multi-band wide dynamic compression), while other features are more associated with fitting of technologic processing to achieve a particular aim (e.g., loudness normalization or equalization, strategies to differentially maximize speech intelligibility vs sound quality). In any event, whether one is attempting to identify the role of an established feature or to evaluate the role of, and candidature for, a new feature, there are two requirements that are paramount. The first of these is a way of understanding, characterizing, and quantifying the needs of a hearing impaired listener. Until one has access to the information that can adequately describe what the deficits are, and when they occur, for hearing-impaired listeners, then no systematic selection of features can take place. Furthermore, when one is attempting to assess the extent to which a particular set of hearing aid features which make up a fitting have been successful or otherwise in overcoming these deficits, then some measure of the reduction in those deficits is going to be required. I contend that no matter how many speech identification procedures any researcher or practitioner indulges in, covering whatever speech material, presentation levels, signal-to-noise ratio, type of noise, orientation of target speaker, and interfering speakers or noise, they never will provide an adequate description of the deficits suffered by hearing-impaired listeners. Rather, the only practicable way to access this information (and it is worth noting that the face validity of this mode of access is immediately apparent to all parties) is via structured questioning of the hearing-impaired person. This is particularly true given the various dimensions of hearing ability and hence disability that can be part and parcel of everyday listening. This point is perhaps worthy of some amplification and development.

It would be an error to make the assumption that what we are always trying to achieve is a maximization of speech intelligibility (under whatever conditions are appropriate to individual hearing-impaired listeners). There is more to listening than simple segmental intelligibility, and there is a body of evidence that suggests that features of a hearing aid fitting that we believe to be most appropriate for maximizing speech intelligibility under one set of listening conditions are not necessarily those that are appropriate for other listening conditions, and furthermore, are not those that are necessarily appropriate for maximizing other qualities of listening such as sound quality or listening comfort (Byrne, 1986; Byrne, 1998; Keidser et al., 1995). Thus until we are able to gain access to the situations that are relevant for hearing-impaired listeners (either as individuals or in an experimental group), and their requirements for those situations, then we are never going to be in a position to match our technologic and rehabilitative options to ultimately manage the deficits that those listeners experience. I know of no experienced audiologist who would ever attempt to treat a hearing-impaired listener without access to a description of the problems they encounter, the environments within which they encounter those problems, and the impact these problems have on the hearing-impaired listener and those around him or her. That is part and parcel of the interaction between the rehabilitative audiologist and the individual with a hearing impairment. What is less clear (and has indeed been questioned by practicing audiologists) is the value of a structured approach (or indeed the restrictions of a structured approach) and the extent to which a structured approach might actively inhibit the normal course of the audiologist-client interaction.

It goes without saying that the self-report information about pre-intervention disability and handicap and the extent to which there are reductions or residual deficits following intervention is of primary interest to both the audiologic practitioner and the client. The individual with hearing impairment must make known to the audiologist the problems that they experience and for which they are seeking assistance, and the audiologist clearly needs access to the information when deciding on the optimum course of management and evaluating its success. There is often a third party in such a scenario and that is the funding agency or service provider who is required to find the resources for any intervention. Although this funding agency will in a private practice environment be the clients themselves, I suggest that in many respects the perspective of the funder still enters the equation and is worthy of consideration.

Irrespective of the mode of health care delivery within any system, some entity still has to find the resources to pay for the service. In the United Kingdom and many other European countries, hearing health care is paid by a state system funded out of general taxation and is free at the point of need and delivery, while in other countries compulsory medical insurance is the norm. In the United States there is a large private-pay market, which interfaces with relatively widespread insurance arrangements and also the Veterans Administration system. Irrespective of these divergences in delivery systems, the individuals or entities who are responsible for the funding of the service are increasingly requiring that service providers demonstrate evidence concerning the clinical effectiveness and cost effectiveness of the services that they offer, and in particular of the options that they have offered and selected for particular individuals and groups. This is one particular example of the ever-growing emphasis on evidence-based health care. I do not propose a treatise on health economics, so will sidestep particular issues of cost effectiveness and concentrate on the ways in which one can demonstrate clinical efficacy and effectiveness.

Given that the objective of our interventions is to reduce disability and handicap, then clearly we have to provide information concerning the extent to which we are able to fulfill those objectives. I contend that no freeform interaction between a practitioner and client, no matter how probing, experienced, and committed that practitioner may be, is ever going to provide the sorts of quantitative information that funding agencies require and are going to increasingly require before reimbursement will take place. There is absolutely no doubt within large scale systems such as the National Health Service in the United Kingdom, that the health authorities responsible for funding services are acutely aware of the heterogeneity that exists among practitioners in the options that they deliver to hearing-impaired listeners, and the subsequent effects that this has on costs. When a service provider attempts to justify a particular element of their service delivery (and this element may not necessarily only be a particular hearing aid hardware or fitting feature, but may constitute the recruitment of a particular type or grade of staff or a particular protocol for the matching of management to need), unless they are able to produce unbiased and robust evidence of the extent to which the options that they choose to make available do contribute to overcoming the patients' problems, then funding streams will not follow. I confidently predict that such requirements will be imposed by third-party insurance agencies and other systems in North America (such as those operated by the Veteran's Administration). I also predict that with the spread of information facilities and resources, hearing-impaired people approaching a private-practice audiologist will demand to see evidence of overall effectiveness of the service that they are considering, and the particular effectiveness of the options that are being offered to them.

I suggest that the requirements and responsibilities that are borne by professionals charged with delivering services make it a future requirement that they gather information in such a way that they can provide quantitative data to guide their practice, to offer and communicate informed options to hearing-impaired listeners, as well as provide robust information concerning the clinical effectiveness and cost effectiveness of their services to those individuals and agencies responsible for resourcing the enterprise.

It is this background that has lead researchers to generate a large number of structured self-report measures with different objectives, styles, and characteristics that in one form or another are all advertised as measurements of potential utility to audiologic practitioners. Any sane practitioner must be bewildered by the array of options that are potentially available, and it therefore comes as little surprise that while outcome measures have been a growth industry for researchers, the extent to which they have been taken up in any routine form by practitioners has been quite limited. There is a distinct technology transfer gap between audiologic researchers and practitioners particularly in this field of hearing aid outcome measures, and for this I suggest both parties should shoulder some responsibility. On one hand, practitioners may have unrealistic requirements and expectations, while audiologic researchers can be unaware of the particular restrictions and requirements that clinical practice imposes.

I was fortunate enough to be at an international meeting recently as part of a workshop to discuss practitioners' requirements for an outcome measure to fulfill at least some of the purposes previously outlined. It was instructive to hear their requirements and the way that they were phrased, these included:

1. That the outcome measure should be sufficiently stable and sensitive to differentiate between realistic hearing aid feature alternatives for an individual with hearing impairment.

2. That the outcome measure should be sufficiently detailed to inform clinical decision making.

3. That the outcome measure data should be collected without interfering in any way with the normal interaction between audiologist and client.

4. That the outcome measure should be administered without active intervention by a professional, for example, as a postal self-completion format.

5. That the outcome measure should be tailored for specific clients while providing quantitative information for service monitoring and quality standards.

One does not need to be a research scientist to glean that such a set of requirements, if not mutually contradictory on occasions, is going to be extremely difficult to satisfy in practice, given the tension between sensitivity, stability, brevity, coverage, and simplicity. This is perhaps akin to demanding a speech identification procedure that can be completed in under 5 minutes, can differentiate between two appropriate hearing aid fittings, and can be representative of real world listening—something that is not likely to be available in our lifetime, if ever.

I have argued elsewhere (Gatehouse, 2000a) that researchers and practitioners cannot specify the requirements for and performance of an outcome measure without a more specific and tighter specification of the purpose of the measure and the environment within which the measure is to be used. I will not reiterate those arguments in detail here. Just as I argue that no statements regarding the utility or validity of a particular measure can be made without consideration of the purpose, population, and setting, I also contend that quoting norms or critical differences without such attention is unhelpful, and indeed potentially dangerous. I will now proceed to review a small subset of the outcome measures produced by the research community and pass on some highly personalized comments on the ways in which they may or may not be of use to practitioners with particular objectives to fulfill. In making these selections I have indulged in a certain amount of personal bias, but have attempted to select those with very different characteristics and those that might be realistic contenders, were a funding agency to insist on some form of routine outcome measure of service monitoring and quality standard purposes. The consideration is (roughly) in chronologic order and reflects the growing expansion and sophistication of what self-report measures purport to measure.

Hearing Handicap Inventory for the Elderly

The Hearing Handicap Inventory for the Elderly (HHIE and its offshoot the Hearing Handicap Inventory for Adults-HHIA) was perhaps the first of the relatively widespread self-report outcome measures (Bentler and Kramer, 2000; Newman et al., 1990; Newman et al., 1991; Noble, 1998; Ventry and Weinstein, 1982; Weinstein et al., 1986) intended for potential routine clinical use. This measure found its way into audiologic research and evaluation, and to some extent clinical enterprises. That is not to say that other more extensive measurements had not been available (Bentler and Kramer, 2000) and subjected to a variety of psyhcometric and other evaluations (Hyde, 2000; Noble, 1998) with regard to stability, coverage, and so on. However, this was a measure which was sufficiently short (with twenty five items, as opposed to the many tens or even some hundreds of items which research measurements might contain), that had a specific configuration toward assessment of disability and handicap and the potential for assessing hearing aid benefits. The HHIE consists of 25 questions directed at hearing disability and handicap, which provides an overall score and two particular subscales. The questionnaire is completed in the unaided state (usually before hearing aid fitting) to provide a pre-fitting score and subscores, and then re-administered after hearing aid fitting to provide a post-fitting score and subscores. The two subscales from the HHIE are labelled emotional and social/situational. Examples of the items are:

1. Emotional: “Does a hearing problem cause you to feel depressed?”

2. Social/situational: “Does a hearing problem cause you to use the phone less than you would like?”

The differences between the pre-fitting scores and the post-fitting scores are measurements of the benefit of intervention, while the extent to which the post-fitting scores may or may not diverge from an appropriate control group measure the extent to which residual deficits remain following the intervention. The HHIE concentrates to a large extent on the emotional and psychological consequences of impaired hearing, and hence the derived measure of benefit focuses on the extent to which those consequences are alleviated by the intervention. Elsewhere I have proposed a criterion (Gatehouse, 1997; Gatehouse, 1999b) that hearing aid outcome measures should satisfy if they are to find use in comparing various technologic options in hearing aid features, as opposed to comparing options in rehabilitative delivery context. I merely note here that the extent to which scores from the HHIE correlate with technical aspects of the hearing aid fitting, speech identification scores, and other outcomes in the disability and handicap domain is somewhat limited. This then would appear to compromise the HHIE as a potential contender in the evaluation of technologic options for hearing aids (Dillon et al., 1997; Gatehouse, 1997) given its relative in-sensitivity to the technical success of a fitting. A fuller investigation of these characteristics may be found in Noble (1998).

The Abbreviated Profile of Hearing Aid Benefit

The Abbreviated Profile of Hearing Aid Benefit (APHAB) is a self-report measurement that is firmly anchored in the disability domain. It is a self-assessment inventory, during which patients report the amount of trouble they are having in various everyday circumstances. This shortened questionnaire was derived from longer predecessors the Profile of Hearing Aid Performance and the Profile of Hearing Aid Benefit (Bentler and Kramer, 2000; Cox and Gilmore, 1990; Cox et al., 1991; Cox and Rivera, 1992; Cox and Alexander, 1995). It produces four subscales as well an overall score, the subscales being labelled; ease of communication (strain of communicating under relatively favorable conditions), reverberation (communicating in reverberant rooms), background noise (communicating in noisy settings), and aversiveness (unpleasantness of environmental sounds). Like the HHIE, the APHAB is administered in an unaided and aided state separately, and a measurement of benefit derived by subtracting the two elements. Positive values for all aided/unaided differences indicate good structure. In practice, difference scores on the aversiveness scale are often negative following amplification. Examples of the items are:

1. Ease of communication: ‘When I am in a small office interview or answering questions, I have difficulty following the conversation.’

2. Reverberation: ‘When I am talking with some one across a large empty room I understand the words.’

3. Background noise: ‘When I am in a crowded grocery store talking with the cashier I can follow the conversation.’

4. Aversiveness: ‘Traffic noises are too loud,’ for each of these statements the listeners are offered a seven point response scale.

The APHAB can be self-completed using a pencil and paper format and is also available in a software format. Robyn Cox has written extensively about the APHAB and the use of both the global score and the profile across subscales: suffice it to say that the putative applications include:

1. The prediction of success from unaided scores (e.g., given the information about pre-fitting problems, what does our experience imply about the likely outcome of a particular management?)

2. The evaluation of the absolute success of a fitting (e.g., what does the post-fitting information tell us about our success or failure in overcoming the listener's problems?)

3. The evaluation of alternative fittings on a patient by comparison of two aided profiles within an individual listener (e.g., can we use the information to determine which hearing aid feature or fitting is best for a hearing impaired listener?)

As with any standardized outcome measure, norms for particular populations can be derived and used as a reference against which experimental groups, clinical practices, or even individual hearing impaired patients can be compared. However, we shall ensure that the norms are appropriate or the purpose, population, and setting.

As indicated, the APHAB is firmly anchored in the disability domain and pays little or no attention to the emotional and psychological consequences of impaired hearing or any of the aspects of service delivery that might affect outcome. Thus one might argue that it is (given the emphasis on speech understanding) more suited than the HHIE (with its emphasis on emotional and social consequences) for the evaluation of particular hearing aid fittings and features that are designed to facilitate understanding speech in particular environments, providing those environments have examples in the set of items within the questionnaire inventory.

The Client-Oriented Scale of Improvement

Both the HHIE and the APHAB demonstrate some of the limitations of fixed inventories. A short inventory such as either of the above is unlikely to contain all of the listening circumstances and consequences of impaired hearing that are relevant for any hearing impaired listener. However, practical inventories developed for potential clinical use have to be short. The more extensive inventories containing some tens or hundreds of items are just not compatible with the clinical practice, although they might be contemplated in intensive research enterprises (Bentler and Kramer, 2000). Thus the trade-off between coverage and resources is something that the fixed inventories cannot resolve; indeed, even if time were available for the longer inventories, this would suffer from the inevitable consequences of asking a large number of irrelevant questions for particular individuals. The client-oriented scale of improvement (COSI) attempts to overcome this limitation by building its total configuration around a client-specified element (Bentler and Kramer, 2000; Dillon et al., 1997; Dillon et al., 1999). There are no fixed listing situations or consequences of impaired hearing in the COSI. At an initial interview the client is asked where and when they would like to be able to hear better, and are asked to indicate up to five circumstances in order of their significance to the individual hearing-impaired person. Listeners are encouraged to be as specific as possible in these nominations. At the time of post-fitting evaluation, they are asked to rate both the degree of change and the final ability on a five-point scale.

Although the COSI was configured primarily as a tool for clinicians and to be as flexible as possible and compatible with clinical practice, it does contain the possibility of quantifying the open-ended responses into scales for both change and final ability. Furthermore, to provide numerical scales one can categorize the specific needs into a set of prespecified groupings for statistical purposes.

Because of the extreme thought put into the COSI to make it as clinically acceptable as possible, it is perhaps the measure with the largest reference base and the one for which practitioner reactions are most readily available (Dillon et al., 1999; Dillon and Mei, 2000).

Although outcome measures such as the HHIE and the APHAB have been aimed at routine clinical practice, they have predominantly been used in research. Almost uniquely, the COSI has a large body of data that has been garnered in a clinical service delivery context (Dillon et al., 1999) and not in a research context. There are a large body of that data on the relationship of the COSI to other outcome measures (Dillon et al., 1997) and on the extent to which the information it provides does or does not correspond with the alternative possibilities. Furthermore large scale surveys of clinicians have identified the factors in the COSI, and by implication in other outcome measures as well, that either do or do not find favor with clinicians (Dillon and Mei, 2000) and hence will influence take-up of the research products that we attempt to recommend. Readers are referred to Harvey Dillon's contribution to the recent Eriksholm Workshop (Cox et al., 2000) on outcome measures for details (Dillon and Mei, 2000). Outcome measures may be summarized as being positive when they provide information that is not available from other sources without impinging overly on the clinical interaction. Indeed, they may facilitate the clinical interaction. Outcome measures are deemed negative when duplication occurs and when the resource commitments outweigh the benefits. Clearly because it is designed to be part of and indeed contribute to the clinical interaction, the COSI cannot be used as an external (after the event) evaluation of a service by an external third party. This is a prime example where the design of an measure to fulfill one particular purpose is in direct contradiction to its use in an alternative context for an alternative purpose (Gatehouse, 2000b).

Before attempting to discuss the potential advantages and disadvantages of the measures discussed so far, I now turn to measures that attempt to capture elements beyond disability and handicap.

Satisfaction with Amplification in Daily Life

The measures that we have dealt with so far attempt to focus on client's problems, disabilities, and handicaps, and have their strengths and weaknesses in those domains. Clearly they are strong contenders when one wishes to evaluate different hearing aids or when different systems avail themselves of different hearing aid features. There is, however, much more to a service, its configuration and delivery, than the technologic content of the hearing aids that it uses. It was in recognition of this complexity that Cox and colleagues configured the satisfaction with amplification in daily life questionnaire (SADL) (Bentler and Kramer, 2000; Cox and Alexander, 1999). Satisfaction is a complex component with many elements, and consequently, the literature on this topic is extensive. To access some of the important elements and limitations, readers are referred to Buhrlen-Armstrong et al., 1998; Chung et al., 1999; Kravitz, 1998; Lambert et al., 1998; Meredith, 1993; Meredith, 1996; Reis et al., 1999; Rey et al., 1999; and Wheelock et al., 1998. The SADL is a self-administered questionnaire designed as a clinical measure of satisfaction from the patient's point of view, though unlike the APHAB it is not administered in a two stage process (corresponding to unaided and unaided). It is administered after the event and the questions are all configured to have an implicit reference. The eventual questionnaire contains 15 items from which can be formed a global score and four subscales. The four subscales are labelled positive effect, service and cost, negative features, and personal image. Examples of the items are:

1. Positive effect: ‘Reduction in how often you ask people to repeat themselves’

2. Service and cost: ‘Competence of the dispenser’

3. Negative features: ‘Feedback when the hearing aid is turned up’

4. Personal image: ‘Does the hearing aid make you seem less capable’

Clearly the constellation of elements is quite unlike the disability orientation of the APHAB and is now tapping into elements of the service and indeed of the hearing aid that are likely to be at least to a degree independent of the extent to which the amplification characteristics of the hearing aid make the audibility of signals and understanding for speech accessible to the listener.

A realistic question for practitioners then to ask is “Why and when should I use the APHAB or HHIE” and “Why and when should I use the SADL.” Once again, this emphasizes the issue of the underlying purpose for which the evaluation is required. If a practitioner wishes to evaluate a newly available technologic feature in hearing aid fitting (let us select for example, a microphone whose directional characteristics may be automatically configured by what the hearing aid perceives as the spatial orientation and sound sources) then it would appear appropriate to select an measurement that contains enough exemplars of difficult listening circumstances where the inferred improvements in signal-to-noise ratio might be accessed and become evident. One might imagine that the APHAB would be a superior instrument to the SADL for this particular purpose. If one however, was faced with the task of evaluating one service that used one particular type of hearing health-care professional with a particular background, against another service that was based on a different mode of organization (whether or not the two services used the same types of hearing aid hardware and fitting protocols), then a case can be made for use of the SADL. Use of the SADL is most appropriate since it attempts to capture the totality of the quality of a service rather than merely the extent to which the service delivers improved listening in adverse listening circumstances. That is not to say that eventually one will not wish to try and titrate the extent to which a good quality service depends on good quality hardware and fitting as opposed to other aspects of the service delivery environment. There are circumstances where one might wish to assess disability, knowing as one does from the literature that satisfaction with a service, be it rehabilitative or surgical, is dependent upon a whole host of issues independent of the technical efficacy of the various constituent elements of that particular service.

Having dealt briefly with four different outcome measures, I shall now spend a little more time on one particular measurement with which I have been involved in developing and propagating. In doing so I am aware of a bias in this article, but crave the indulgence of the reader in attempting to bring to your attention some of the ways in which the information might be used and examples of its use that, to my knowledge, are not available in the currently published literature.

Design of the Glasgow Hearing Aid Benefit Profile

In considering the desirable properties of an outcome measure, I suggest that, in addition to the required psychometric abilities (Hyde, 2000), an outcome measure should recognize and appropriately assess the different dimensions of disability and benefit. It should retain the ability to address the relevant concerns of individual hearing impaired listeners while providing outputs that are analytically tractable and psychometrically viable. I argue that good audiologic practice will require, in addition to some measure of auditory impairment prior to hearing aid fitting, an assessment and recognition of the disability experienced by the client, and the extent to which that disability impacts upon the client's life (hearing handicap). When an audiologist wishes to ascertain the extent to which intervention has been successful, it will be required to have some indication of the extent to which the client makes use of the hearing aid, the extent to which a hearing aid fitting has delivered benefit to the client, the extent to which the client has residual problems that may or may not require further intervention, and the extent to which the client is satisfied by the overall delivery of the service. The Glasgow hearing aid benefit profile (GHABP) attempts to access these dimensions while allowing the client to participate in the problem setting and solving exercise by determining the goals of the intervention.

Space in this article does not permit a full description of the steps and principles that have entered into the generation and validation of the GHABP; details are available elsewhere (Gatehouse, 1997; 1998; 1999a; 1999b; 1999c; 1999d; 1999e; 2000a). At the outset, a pragmatic requirement was constructed in that we require that an outcome measure compatible with clinical practice should be printed on both sides of a single A4 page. Figure 1A shows the front page, which contains four prespecified listening circumstances. The first stage is to determine whether a listening circumstance occurs in a listener's lifestyle; if it does not, the six columns or dimensions are ignored—this avoids spending time and effort on matters irrelevant to the client. When a circumstance occurs, the degree of difficulty is assessed (initial disability) and also the impact on the client's life (handicap). The first two columns are completed prior to management.

Figure 1.

The Glasgow hearing aid benefit profile (GHABP). A, represents the first page of the GHABP containing the four prespecified listening circumstances, while B represents the reverse side and contains the (up to) four listener-specified listening circumstances.

We have dealt with some of the situations which in our experience can lead to difficulty with hearing. What we would now like you to do is to nominate up to four new situations in which it is important for you as an individual to be able to hear as well as possible.

Following intervention, the extent to which the hearing aid is used (use), the extent to which problems have been reduced (benefit), the extent to which problems remain (residual disability), and the extent to which the client has been satisfied with the intervention, are ascertained for each of the four listening circumstances. Hearing aid use is assessed in terms of the proportion of the time that clients make use of the intervention in environments that exist in their lifestyle and cause them difficulty.

The individual relevance of the profile is achieved on the reverse side of the A4 page (Figure 1B), where clients are invited to specify (up to) four listening circumstances for which it is important for them to be able to hear well in their everyday existence. Table 1 shows the distribution of the listener-specified circumstances in the GHABP for 293 patients at Glasgow Royal Infirmary. For example, “listening in church” was specified by 14.3% as their first open-ended element, by 11.6% as their second, by 5.5% as their third, and was never specified as the fourth open-ended choice. All listeners specified at least one circumstance, 89.8% at least two, 80.2% at least three, and 65.9% specified four. Analyses have shown that the psychometric power (in addition to the rehabilitative relevance) of the GHABP arises in large measure from the listener-specified elements (Gatehouse, 1997; Gatehouse, 1999b). A detailed manual for the administration and scoring of the GHABP is available (Gatehouse, 1999a) as is a software version of the questionnaire. At this stage it should be stressed that the GHABP is not a questionnaire that is given to the client for them to complete on their own during, before, or after the management process. It is configured to be part and parcel of the rehabilitation intervention and to be completed by interview. It is argued that the GHABP provides a formalization of what would be regarded in any circumstance as good audiologic practice.

Table 1.

Distribution of the Subject-Specified Situations in the GHABP

Situation	1	2	3	4
Listening in a pub or club	16.4%	7.8%	2.4%	2.7%
Listening on the telephone	14.3%	11.6%	5.5%	3.8%
Listening in church	12.3%	9.6%	11.3%	0%
Listening to young children	9.9%	8.5%	4.4%	4.1%
Listening in a bus or car	8.9%	2.4%	5.1%	2.7%
Listening in a meeting	6.8%	10.9%	8.9%	1.0%
Listening at bingo	6.5%	0.7%	2.7%	0.7%
Telephone ringing	3.8%	3.1%	0.3%	1.4%
Listening at work	3.8%	1.4%	0.7%	2.4%
Doorbell	2.7%	4.4%	1.4%	1.0%
Listening in the cinema	2.7%	3.1%	2.7%	1.4%
Listening to tannoy announcements	2.7%	1.0%	3.1%	1.4%
Determining the direction of a sound	1.7%	2.7%	0%	2.7%
Listening to radio and music	1.4%	3.4%	3.1%	3.8%
Listening to a lecture	1.0%	1.7%	1.4%	4.4%
Hearing a voice from another room	1.0%	0%	0.3%	2.7%
Listening at the theatre	0.7%	5.8%	7.8%	4.4%
Exercise class/swimming pool	0.7%	1.4%	1.7%	5.1%
Noise when sleeping	0.7%	1.0%	1.0%	4.1%
Conversing without lip-reading	0.7%	0.3%	0.3%	2.4%
Listening to an accent	0.3%	2.4%	3.8%	4.1%
Social functions with background music	0.3%	2.0%	1.7%	3.8%
Watching TV alone	0.3%	1.7%	2.4%	1.0%
Family gatherings	0.3%	1.0%	5.8%	2.7%
Other	0%	1.7%	2.4%	2.0%
Not used	0%	10.2%	19.8%	34.1%

The GHABP furnishes six scales, each of which is configured to vary between 0 and 100. Zero on the scale of initial disability corresponds to having no difficulties in any of the circumstances occurring in the patient's lifestyle and for which it is important for them to be able to hear well, while 100 on the scale of initial disability corresponds to “cannot manage at all” in those circumstances. Zero on the scale of handicap similarly corresponds to having no impact, while 100 corresponds to being worried, annoyed or upset to an extent labelled as “very much indeed.” A post-intervention hearing aid use score of zero corresponds to never using the hearing aid at all in any of the circumstances that exist in a client's lifestyle, are important to be able to hear well in, and prior to fitting led to a hearing difficulty. A hearing aid use score of 100 corresponds to using the hearing aid “all of the time” in all of those circumstances. It is argued that this scale of use (as opposed to absolute hours per day) provides a metric that is more amenable and tractable and appropriately scaled to a client's needs. A reported hearing aid benefit score of zero corresponds to the hearing aid being “no use at all” in any of the circumstances that exist in a client's lifestyle, are important to be able to hear well in, and led to a hearing difficulty prior to intervention. A score of 100 corresponds to hearing perfectly with the hearing aid in all of those circumstances. Similar scale correspondences are derived for residual disability and satisfaction. The psychometric evaluation and validation of the GHABP is described in detail elsewhere (Gatehouse, 1997; Gatehouse, 1999b) and in this article we concentrate on the demonstration of the ways in which data from the GHABP might be used to evaluate aspects of a system for service and are of potential utility in the treatment of individuals with hearing impairments.

In clinical settings there are a wide range of technologic hearing aid options available to practitioners. Therefore a useful outcome measure should be appropriately sensitive to changes (which are hopefully improvements) in the technology available and the prescription or adjustment regimes that are used to match the technologic options to the characteristics of hearing impaired listeners. As an example I will show some data for a particular context, the National Health Service (NHS) in the United Kingdom (UK). I appreciate that this form of service delivery is peculiar to the UK, but the point I wish to present is more general. Although no comprehensive UK data are available on the impact of advanced technology (indeed, one of the intentions in providing measurements such as the GHABP is that it can be used as a vehicle to generate such information), the context of the UK NHS might be used to demonstrate the extent to which the GHABP is sensitive to changes in technology. The UK NHS utilizes predominantly linear post-aural hearing aids in contrast to the widespread implementation of in-the-ear devices in most developed countries, leading to public and professional pressure for their adoption in the United Kingdom. The National Health Service in Scotland sponsored a trial of modular in-the-ear devices that took place in clinics in Ayrshire, Aberdeen and Glasgow. Figure 2 shows the data from the GHABP scales of reported use, benefit, and satisfaction. Also shown are audibility values (the speech intelligibility index) (ANSI, 1993; Berger, 1990), introduced by the different devices separately for standard post-aural fittings and in-the-ear fittings. Each panel in Figure 2 shows the cumulative distribution of the scale variable. Thus in the top left panel reporting “use”, 10% of BTE fittings have scores less than 10 and 10% of ITE fittings have scores less than 40. The values for the 50th percentile (median) are approximately 68 and 78, respectively. The point of including the data here is not to provide a case for or against the clinical effectiveness and cost effectiveness of modular ITE as opposed to post-aural technology in the UK NHS, but rather to demonstrate the ways in which the different scales of the GHABP might be used to evaluate any benefits that might occur. It can be seen from the speech intelligibility index (SU) data in Figure 2 that the ITE fittings lead to greater aided audibility of a speech signal than do the post-aural fittings. This is due to the increased high-frequency output of the ITE devices following the removal of earmould effects. Such improvements in high-frequency audibility are shown to be accompanied by improvements in speech intelligibility in noise. The three scales of the GHABP demonstrate the extent to which the ITE fittings are accompanied by increases in hearing aid use, hearing aid benefit, and satisfaction with the hearing aid fitting and can be substantial and occur throughout a wide range of the distribution of outcomes on those scales. Thus, system data such as that shown in Figure 2 can be used to access the extent to which different technology and fitting procedures are accompanied by changes in the self-report domain. These can then be offset against the potentially increased costs of the new provision strategies.

Figure 2.

The distribution from the GHABP of reported use, reported benefit, and reported satisfaction and aided audibility from the speech intelligibility index (SII) separately for the modular in-the-ear fittings (solid lines) and post-aural fittings (dashed lines).

While the data in Figure 2 show that the GHABP is sensitive to technologic changes, these are not the only components of service development and optimization that might be available. An equally important component is the rehabilitative context within which the hearing aid fitting occurs. Therefore, it is of importance and interest to be able to document the effects of and potential candidature for such rehabilitative effort. This article uses as an illustrative example the scheme of pre-fitting counselling (Brooks and Johnston, 1981; Norman et al., 1998) previously operated at Withington Hospital in Manchester whereby clients were visited in their homes prior to hearing aid fitting. Although there is some evidence about the effectiveness of this intervention (Brooks and Johnston, 1981), other work draws the opposite conclusion (Norman et al., 1998). As part of the collaborative effort that led to the development of the GHABP, data are available from two matched samples at Withington Hospital, Manchester, who received the same standard post-aural hearing aids, but one subset received the pre-fitting counselling and other received the standard provision environment. The data from the GHABP from the two groups are shown in Figure 3. Here the improvements in outcome are less in magnitude and less sustained throughout the distribution than those from the different technologic options in Figure 2. Briefly the effect of pre-fitting counselling can be seen to provide an improvement in the outcome for those listeners for whom the prognosis was poor; that is, improvements in use, benefit, and satisfaction can be demonstrated for the lower ends of the distribution. As with the previous example, the point of including the data in this article is not to argue for or against the clinical effectiveness or cost effectiveness of pre-fitting counselling, but rather to act as a demonstration of the extent to which the GHABP can provide useful scales that are sensitive to different components of the intervention.

Figure 3.

The distribution from the GHABP of reported use, reported benefit, and satisfaction separately for the patients at Withington Hospital, Manchester, who received pre-fitting counselling (solid lines) and standard provision (dashed lines).

While such demonstrations are important, if an outcome measure is to find acceptance by professionals and hearing-impaired patients, the ability to demonstrate utility at a system level is likely to be insufficient. The GHABP has been configured as a potentially useful tool to the audiologist in the overall rehabilitation process. The scores from the GHABP can be useful in their own right when some form of gold standard is adopted. For example, it might be adopted as a goal for an individual with hearing impairment to use a hearing aid all of the time in all of the circumstances that exist in his or her life, are important for them to listen to, and caused them difficulty prior to intervention; and to be able to hear perfectly with the hearing aid in all such circumstances. However, given the state of current hearing aid options and rehabilitation, this is perhaps an unrealistic goal, though one for which prospectively collected information about the extent to which the goal has been reached is of prime importance. Of perhaps more relevance to individual hearing-impaired patients and audiologists confronted with the task of providing treatment for an individual, is the extent to which provision of a hearing aid has provided outcomes comparable to those of peers in the clinical population. That is, given the characteristics of the patient (and these might include characteristics of the auditory impairment, age, etc), is the client doing better or worse than one might have expected given our experience with the technology and rehabilitation at our disposal in our own clinical circumstances. Thus when a service has been collecting data on the GHABP for some time, they will have a reference against which to judge the performance of individual hearing-impaired listeners. Thus, any value on each scale on the GHABP can be converted into a percentile value of the clinical population peers of the individual client. Figure 4 shows one such representation from a software implementation of the GHABP. The six scales of the GHABP are represented horizontally. Each of the scales is configured so that poorer performance is towards the left of the figure. Thus high initial disability, high handicap, low use, low benefit, high residual disability, and low satisfaction each represent poor performance. The “button” on the scale and the figure in the text box to the right of the scale represent the location on the percentile distribution of clinical peers of this individual. Thus the example in Figure 4 resides at the 48th percentile of initial disability, the 81st percentile of handicap, the 61st percentile of use, the 21st percentile of benefit, the 86th percentile of residual disability, and the 23rd percentile of satisfaction. This example then demonstrates close to median initial disability, high handicap, above median use, low benefit, high residual disability, and low satisfaction. This example is used to try and demonstrate the ways in which this profile of results might guide future management decisions. If such an individual has been treated with a single post-aural peak-clipping hearing aid, an audiologist might be tempted to overcome the low benefit and high residual disability by offering more complex and expensive technologic options. The argument here however, is that this listener suffers from a mismatch between disability and handicap (the individual is much more handicapped by their hearing disability than one would expect from equivalent data from peers in the clinical population) and therefore until the origin of that mismatch is identified and probed via the rehabilitation process, the expenditure of either the state's or the patient's resources on more expensive hearing aids is unlikely to be productive.

Figure 4.

Profile of the location on the percentile distribution of fittings in Glasgow and Manchester on the seven GHABP scales for clinical example A.

This profile is contrasted with that shown in Figure 5, where there is a reasonable match between initial disability and handicap, high hearing aid use but low hearing aid benefit, high residual disability and low satisfaction. In this example, the patient is making good use of their hearing aid, but despite this is receiving low benefit and suffering high residual disability. It can be argued that such a profile demonstrates appropriate candidature for more intensive signal processing options. The previous two instances are specific examples of the ways in which the scale profiles might be used to further individual patient treatment, in addition to furnishing the system information discussed earlier.

Figure 5.

Profile of the location on the percentile distribution of fittings in Glasgow and Manchester on the seven GHABP scales for clinical example B.

In indulging myself in an extended description of the GHABP and its use and implementation, it has not been my intent to use this article as an extended advertisement. Rather I attempt to identify and highlight some of the design considerations and choices that are open to researchers and practitioners, and then to give examples (naturally from a UK context, but which I argue can be generalized in principle) of the ways in which the information may be put to use.

The 1999 Eriksholm Workshop on Outcome Measures

I have been acutely aware of the ear and hearing supplement that constituted the output of the 1999 Eriksholm workshop of the Oticon foundation on the topic of Self-Report Outcome Measure in Audiological Rehabilitation including Hearing Aids. I have not duplicated much of the information and arguments contained in that supplement, but direct the readers' attention to it because it reviews the literature in a much wider and more extensive manner and presents a set of perspectives that differ from mine (Cox et al., 2000). By including it as a subheading, I attempt to stress a number of issues that are contained therein but are not necessarily conventional elements of the audiologic literature. I have already referred to the health care scenarios that drive the agenda, and references within the supplement deal with these in greater detail and from a perspective more detached from audiologic practice (Beck, 2000; Wilkerson, 2000; CARF, 1997; 1998; 2000; Fuhrer, 1987; Gagne, 2000; Kricos, 2000).

Quality of Life

Given their relative unfamiliarity to audiologic audiences I intend to concentrate on the issue of health-related quality-of-life measures and the role that such measures may play in audiologic practice. A generic measure is one that by definition refers to no particular condition or consequences of that condition. Rather it attempts to assess those elements of well being that might be affected by a disease process. When might one then require or even desire to use such a measure in preference to the condition-specific disability and handicap measures that we have been considering? This requirement is likely to occur when we are attempting to engage funding organizations (and these are usually at the level of authorities and governments rather than the individual; note that this is an issue of funding for our services as opposed to funding within our services). When one is attempting to engage the debate concerning the relative clinical effectiveness and cost effectiveness of services for hearing-impaired listeners as opposed to services for clients or patients with other health conditions, then we need to use a metric that is common across the conditions: here we are faced with a dilemma. Almost all the existing generic measures of health-related quality of life were constructed with either life-threatening diseases as their target or conditions that had material effects on mobility. As such, they assess in an imperfect manner the social, emotional, and physiologic consequences of conditions, particularly chronic conditions, such as sensory disabilities. Thus we are faced with a playing field that is anything but level. We do though have to convince the health services research and health economic communities that there is an issue to be addressed if an equitable prioritization of allocation of health care resources is to be achieved. Clearly this is not an issue for audiologic practitioners, but rather for the research community. Examinations of the strengths and weaknesses of the existing generic measurements and their uses and misuses forms part of the product of the Eriksholm Workshop (Bess, 2000; Dittmar and Gresham, 1997; EuroQOL Group, 1990; Gilson et al., 1975; Hyde, 2000; McHorney et al., 1992; Torrance, 1986; Torrance and Feeny, 1989). My objective in raising the issue as a specific element in this manuscript is to serve as a warning to practitioners against using generic measures of health-related quality of life. While they show apparent applicability and ease of use, their construction will automatically lead to the conclusion that the conditions with which we encounter and the intervention with which we treat those conditions achieve low scores on the various utility scales that are derived.

As an illustrative example, the EuroQOL questionnaire (as opposed to thermometer) contains five sections:

1. Mobility

   I have no problems walking about.

   I have some problems in walking around.

   I am confined to bed.

2. Self-Care

   I have no problems with self care.

   I have some problems washing or dressing myself.

   I am unable to wash or dress myself.

3. Usual Activities

   I have no problems with performing my usual activities, (e. work, study, housework, family or leisure activities).

   I have some problems in performing my usual activities.

   I am unable to perform my usual activities.

4. Pain/Discomfort

   I have no pain or discomfort.

   I have moderate pain or discomfort.

   I have extreme pain or discomfort.

5. Anxiety/Depression

   I am not anxious or depressed.

   I am moderately anxious or depressed.

   I am extremely anxious or depressed.

Such questions are very unlikely to access the psychological, social, and emotional consequences of a chronic sensory impairment.

The use of such generic measures inappropriately (and particularly to compare alternative technologies at our disposal) will do us and our colleagues no service at all. Generic measures should be used only when one wishes to compare across conditions and interventions for those conditions in disparate health care domains.

Some Issues and Choices for Practitioners

I am sure that readers will be aware by now that I am not offering solutions but rather criteria upon which choices and solutions may be identified. If as a reader you accept my thesis that society is going to increasingly demand a demonstration of the effectiveness of what we do and the extent to which that effectiveness can be aligned with the costs of our services (by whomever they may be borne), we must start to gather that information as part of our routine. There is still a range of options that are open that are likely to vary depending on the particular organization, mode of service delivery, and objectives of the enterprise. What I do suggest is clear: that the list of ‘ideal requirements’ that I put forward earlier is unlikely to be captured within a single measure and therefore some focused choices and compromises have to be made. Practitioners and the customers who those practitioners are attempting to satisfy by the enterprise will need to negotiate the dimensions on which the outcome is required. If an overall evaluation is required, then at least some of the elements that are tapped into by the SADL and to some extent the GHABP will be prime requirements. If the objective includes a systematic evaluation of, for example, introduction of the routine application and routine use of digital technology in the Veterans Administration and its subsequent effects on clinical effectiveness and cost effectiveness, then perhaps more of the elements of the APHAB will need to be included (and note here I am not making a case either for or against any particular measurement rather the objective is to try and bring forward the argument that choices of outcome measurements can only be made in the context of the requirements). Other components that will enter into the set of choices will of course include the time that is available (which will influence directly the length and complexity of any item) and whether the measurement is designed to be a tool that aids the clinician, in which case they will have to be actively involved in the administration and interpretation, or whether a more detached view is required for service monitoring and assessment to reduce clinician effects to lower levels.

If I have one thesis that I attempt to put forward, it is that any measure that is chosen has to be shown to be ‘fit for purpose.’ Thus, if the requirement is to differentiate between services, the researchers who have proposed the measure should have generated data that convinces the potential user that the measure has the required sensitivity. In a similar manner if the requirement is to differentiate between different types of hearing aid fittings and features, then the researchers who have produced various self-report outcome measures should have produced data concerning validity and sensitivity for that purpose. The construction and implementation of the GHABP has indeed contained such criteria. I urge my colleagues in the research community to follow suit.

I do expect to see in the next few years large-scale data sets that will probably initially be derived from large-scale health delivery systems such as the state-funded health-care systems in parts of Europe, the state voucher systems in Australia, and perhaps the Veterans Administration in North America. I am sure that the umbrella organizations responsible for the funding and configuration of these services will increasingly demand that consistent coherent data concerning how those services perform and how they perform differentially in different facilities will be information that will reach the public domain.

It is unlikely that each and every funding agency will have the same purpose and the same requirements of evaluations; therefore, it is likely that each will choose diverse sets of outcome measures for those enterprises. With this in mind that the 1999 Eriksholm workshop constructed the international outcomes inventory for hearing aids (IOI-HA) (Cox et al., 2000). This contained a number of elements that the participants thought were invariant across culture, across service delivery modes, and across the objective of the evaluation. They do not form a replacement for many of the competitor measurements but rather a common currency that any evaluation might include, thereby facilitating cross comparison of the various systems. This is at the early stage of routine outcomes assessment, perhaps a matter for researchers rather than practitioners, but the availability of such a common currency should be brought to the attention of the funding agencies and regulatory authorities that might sponsor any such widespread evaluation exercise.

I hope that I have managed to use this article as a vehicle to bring to the attention of the readers some of the requirements and options that might be at their disposal and some of the ways in which practice might have to be configured in the future to meet the demands of an evidence-based approach to health care, which is going to become more and more prevalent throughout the world. I remain convinced that routine quality standards in audiologic practice will become the norm. In addition to whatever measure one requires to specify technical services of hearing aid fitting, those standards should and will contain elements in the self-report domain. We need to engage funding agencies, researchers, and practitioners in a debate to converge on a set of requirements and specifications to best serve the professional needs of audiology and, of paramount importance, the hearing-impaired people who we serve.